Fault Diagnosis of Industrial Process Based On KICA and LSSVM

The paper combines kernel independent component analysis for establishing the fault detection model and the least squares support vector machine for establishing the fault diagnosis model to set up the industrial process monitor model as the growing difficult for the complex industrial process monitoring. It uses the data collected by the industrial process to extract the nonlinear independent component for establishing the detection model, and put the data into the model of LSSVM to identify the fault only when the fault occurs. Finally, using the data of TE process verifies the validity and practical of the method.